cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
418
Views
0
Helpful
2
Replies

ACS SE 4.2 RADIUS protocol problem

Timelessness
Level 1
Level 1

I am using the Self RADIUS server in my Cisco ACS SE 4.2 appliance S. I have an AAA client C that interacts with S by means of the RADIUS protocol. This works fine, in that S correctly carries out authentication chores on username/password (PAP and CHAP) pairs received from C, sending back to C the corresponding Access-Accept packet when the authentication succeeds, or Access-Reject when it doesn't.

I have been able to import a set of three VSAs into S. Each of those attributes is of string type. I then configured in S a single user U with password P so that, whenever a U/P pair received in S from C is authenticated by S, S should send back to C, in the Access-Accept packet, the three attributes with the following values:

Attribute #1: abc

Attribute #2: xyz

Attribute #3: 123

With this setup, when an authentication is successfully completed by S, C receives 53 bytes worth of data from S every time. I am attaching a typical example, already disassembled. I have disguised the actual vendor ID, for legal reasons, but the rest is exactly as it was when received in C.

According to the disassembly, what we got is an Access-Accept packet, as expected. Its length is 53 bytes - again as expected, for this is the only packet that C has received from S here. However, the packet is incomplete, for attribute #3 is missing its value field.

Looking into the whole packet in more detail, it can be seen that while the wire format for the first attribute, namely, Frame-IP-Address, is correctly constructed, the remaining are not. For example, the sequence of bytes corresponding to the attribute #1 reads 1a 09 00 00 xx xx 2c 61 62 63. I believe that this is incorrect; it should be 1a 0a 00 00 xx xx 2c 61 62 63, for the wire format for this attribute consists of 10, not 9, bytes. I tried a few variations on the values for the attributes, and the results are always substantially the same, in that the wire formats for these attributes are always incorrect.

This all probably implies I have done something wrong when importing the VSAs into S, and/or when configuring things on S. I am therefore attaching the csv files I used to import my VSAs into S; as before, names and vendor ID are disguised, but their lengths are exactly the same as in the undisguised file. I used two csv files: One to import the vendor ID, and the other to import the VSAs under that vendor ID. As for user U, in S's administration GUI I clicked on User Setup and selected user U, moved to the bottom of the screen, where the attributes for this particular vendor were present, introduced the values for each attribute mentioned above, and made sure that button in front of each attribute was ticked.

Any help on this would be much appreciated.

2 Replies 2

Timelessness
Level 1
Level 1

Nobody?

Timelessness
Level 1
Level 1

I have been doing some investigation on patches released by Cisco, and I noticed that one of the defects (CSCsv65072) patched in 4.2.1 seems to be the one I am pointing out. Could someone in this forum please confirm this? Thanks.